Dear all,
I have a panel FE model, dependent variable (Y) is measured in % and also my independent variable (X) is also measured in %.
Coefficient of X= -0.107.
As it is mentioned in a previous post https://www.statalist.org/forums/for...centage-points , the interpretation of coefficient is "a 1 percentage point increase in X results in a 0.107 percentage point decrease in Y".
(I also consider the interpretation using "one standard deviation increase" following some other posts, e.g. https://www.statalist.org/forums/for...20#post1296120 , https://www.statalist.org/forums/for...ard-deviations)
Given all the above, a colleague suggest me that this is not actually true. He said that the above interpretation (first post) does not yield to a percentage point effect (he didn't suggest any other method), but he argue that the percent effect (!percent! not percentage point) is calculated as the ratio of the difference of the linear prediction of X at its mean+st.dev minus the linear prediction of X at its mean divided by the linear prediction of X at its mean (and multiplied by 100).
The formula in STATA (mean of X: 51.777 , st.dev of X: 14.644 , mean+s.d.: 71.821):
Code:
xtreg x y `varlist', fe ro
margins, dydx() predict(xb) noestimcheck force post nose at(X=71.821) //: a
xtreg x y `varlist', fe ro
margins, dydx() predict(xb) noestimcheck force post nose at(X=51.777) //: b

%effect= ((a-b)/b)*100
​​

Any thoughts on this...?

Thank you in advance!